Connecting Ifá Divination Architecture with Machine Learning Prediction Pipelines: Explicit Calculations, Full Derivations, and a Hybrid Framework
Year: 2026
Type: preprint
Abstract: This paper presents a complete mathematical correspondence between the Yoruba Ifá knowledge system and the standard supervised machine-learning prediction pipeline. All central calculations are written explicitly: the derivation that eight binary positions produce exactly 256 odù, the expanded forms of mean-squared error and binary cross-entropy together with their gradients, the layer-wise equations of backpropagation, the formula for Expected Calibration Error, and the routing, retrieval and attestation equations of a hybrid architecture. The exposition is self-contained, implementation-ready, and prepared for publication. It demonstrates how discrete addressing and mandatory interpretation—long present in Ifá—can be combined with gradient-based optimization to produce more auditable and accountable predictive systems.
Author: Mansur O, Abdulrahman
Institution: Federal University Dutsin-Ma
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7236674629
